| Resolver | Median | Best | Worst | Reading |
|---|---|---|---|---|
Cloudflare 1.1.1.1 | – | – | – | waiting |
Google 8.8.8.8 | – | – | – | waiting |
AliDNS 223.5.5.5 | – | – | – | waiting |
How this DNS test works
For each resolver, your browser sends 5 DNS-over-HTTPS JSON lookups for random subdomains and times each response. We report median (typical), best and worst. This measures resolver latency from your connection — the number that decides how fast new domains resolve if you switch to that resolver. It does not read which DNS server your OS currently uses (browsers cannot see that; see the DNS leak test for that side of the story).
Why DNS speed matters (and when it doesn't)
Every visit to a domain your device hasn't cached starts with a lookup. With a slow or distant resolver that's 100–300 ms of dead time before the connection even opens — it reads as “the web feels slow today” even on a gigabit line. Once the lookup is done, DNS is out of the picture: downloads run at full speed regardless. So DNS speed is a snappiness upgrade, not a bandwidth one — and it's free. Two things make it feel even faster: your OS caches answers (repeat visits skip the lookup), and browsers pre-resolve links you're likely to click.
Choosing a resolver
| Resolver | Strengths | Considerations |
|---|---|---|
| Cloudflare 1.1.1.1 | Fast anycast, privacy-first policy (24h log purge), free family filter tier | Run by a CDN — some see a conflict of interest |
| Google 8.8.8.8 | Broadest anycast coverage on earth, very reliable | Google-scale telemetry; DNSSEC validated |
| AliDNS 223.5.5.5 | Fastest for mainland-China access, domestic CDNs resolve correctly | China jurisdiction for logs |
| Quad9 9.9.9.9 | Malware-domain blocking built in, Swiss non-profit | Blocking adds a small share of false positives |
There is no universal winner: anycast means the “closest” server differs by country and ISP, which is why a live test from your own line beats every blog comparison table — including this one.
How to change your DNS
- Router (best — covers the whole house): in OpenWrt, Network → Interfaces → WAN →
Physical/DNS settings, or better, install DoH via
https-dns-proxyso lookups are also encrypted from the router. See our OpenWrt tuning guide for where DNS settings fit alongside SQM. - Windows: Settings → Network → Adapter options → IPv4 properties → “Use the following DNS server addresses”.
- macOS: System Settings → Network → Details → DNS.
- Android: Private DNS (DoT) by hostname —
one.one.one.oneordns.google— cleaner than per-network IPv4 edits. - iOS: install a DNS profile or use an app; per-network DNS is buried in Wi-Fi details.
Plain DNS (port 53) is visible to your ISP — they can log every domain you look up. DoH (DNS over HTTPS) and DoT (over TLS) encrypt that hop. Most modern browsers also offer “secure DNS” in settings, which bypasses the OS resolver entirely.
Related tools
If you're tuning your connection end to end: the bufferbloat test (lag under load), ping test, jitter test and DNS leak test round out the picture.
FAQ
What is a good DNS response time?
Under 20 ms to your resolver is excellent, 20–50 ms is fine for everyday browsing, 50–100 ms makes first visits to new domains feel sluggish, and above 100 ms is worth fixing. The number that matters is from your device to the resolver — a resolver far from you can be fast for local users and slow for you, which is exactly why you should test from your own connection.
Should I use 1.1.1.1 or 8.8.8.8?
Both are excellent, and the honest answer is: whichever is faster from your network, which differs by country and ISP. Cloudflare (1.1.1.1) markets privacy-first logging (24h purge), Google (8.8.8.8) has the broadest anycast coverage. Users in mainland China should consider AliDNS 223.5.5.5 for domestic resolution speed and reliability. Run the comparison above and pick your winner.
Does DNS speed affect download speed?
No — DNS only affects how long the first lookup takes before a connection starts. Once the connection is established, throughput is a bandwidth question. Slow DNS shows up as “the first click on a new site feels slow”, not as slower downloads. Browsers and operating systems also cache results, so repeat visits skip the lookup entirely.
Is this DNS test measuring my current DNS server?
No — it measures how fast public resolvers would be if you switched to them, from your current connection. Browsers cannot read which resolver the operating system uses. To see your active resolver, a DNS leak test observes which server actually answers your lookups from the authoritative side.